New A simplified cell-based assay to identify coronavirus 3CL … · 2020. 8. 29. · 4 60...
Transcript of New A simplified cell-based assay to identify coronavirus 3CL … · 2020. 8. 29. · 4 60...
1
A simplified cell-based assay to identify coronavirus 3CL protease inhibitors 1
2
Samuel J. Resnick1,2, Sho Iketani3,4, Seo Jung Hong1, Arie Zask5, Hengrui Liu6, Sungsoo Kim1, 3
Schuyler Melore1, Manoj S. Nair3, Yaoxing Huang3, Nicholas E.S. Tay6, Tomislav Rovis6, Hee Won 4
Yang1, Brent R. Stockwell5,6*, David D. Ho3*, Alejandro Chavez1* 5 6 1 Department of Pathology and Cell Biology, Columbia University Irving Medical Center, New York, 7
NY, 10032, USA 8 2 Medical Scientist Training Program, Columbia University Irving Medical Center, New York, NY, 9
10032, USA 10 3 Aaron Diamond AIDS Research Center, Columbia University Irving Medical Center, New York, NY, 11
10032, USA 12 4 Department of Microbiology and Immunology, Columbia University Irving Medical Center, New 13
York, NY, 10032, USA 14 5 Department of Biological Sciences, Columbia University, New York, NY, 10027, USA 15 6 Department of Chemistry, Columbia University, New York, NY, 10027, USA 16 17
*Correspondence to: [email protected]; [email protected]; 18
20
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
2
Abstract 21
We describe a mammalian cell-based assay capable of identifying coronavirus 3CL protease 22
(3CLpro) inhibitors without requiring the use of live virus. By enabling the facile testing of compounds 23
across a range of coronavirus 3CLpro enzymes, including the one from SARS-CoV-2, we are able to 24
quickly identify compounds with broad or narrow spectra of activity. We further demonstrate the 25
utility of our approach by performing a curated compound screen along with structure-activity 26
profiling of a series of small molecules to identify compounds with antiviral activity. Throughout these 27
studies, we observed concordance between data emerging from this assay and from live virus 28
assays. By democratizing the testing of 3CL inhibitors to enable screening in the majority of 29
laboratories rather than the few with extensive biosafety infrastructure, we hope to expedite the 30
search for coronavirus 3CL protease inhibitors, to address the current epidemic and future ones that 31
will inevitably arise. 32
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
3
Introduction 33
The outbreak of a novel coronavirus (SARS-CoV-2) infection of the past several months has 34
paralyzed countries around the world1,2. This crisis is further exacerbated by the dearth of approved 35
therapeutics, leaving physicians with few proven treatment options. In the past two decades, the 36
world has already suffered from two other major coronavirus outbreaks, Severe Acute Respiratory 37
Syndrome (SARS) and Middle East Respiratory Syndrome (MERS), suggesting that coronaviruses 38
represent a real and ever-present threat to global health that must be addressed3. Yet, even if 39
therapeutics against the existing epidemic strains are identified, there are several hundred other 40
coronaviruses in active circulation within animal populations, many with the theoretical potential to 41
infect humans. To help identify therapeutics for the current epidemic along with preparing for the 42
next, there is a need for readily deployable small molecule screening assays that enable the 43
identification of therapeutics that are broad-acting across a large collection of coronavirus strains. 44
45
During coronavirus infection, the RNA genome is delivered into cells and translated into a pair of 46
polyproteins4. These polyproteins are then processed by a set of virally encoded proteases, of which 47
the three-chymotrypsin-like protease (3CLpro) performs the majority of cleavage events4. As a result 48
of its essential role in viral replication and high degree of conservation across all coronaviruses, 49
3CLpro enzymes represent important targets for therapeutic drug development5,6. Previous work 50
expressing a variety of viral proteases within yeast and mammalian cellular systems have shown 51
that protease expression can lead to profound cellular toxicity, which can be rescued by the addition 52
of protease inhibitors7–12. We hypothesized that if the expression of coronavirus 3CLpro enzymes 53
within mammalian cells leads to a similar toxic phenotype, this could form the basis of an easily 54
implemented mammalian cell-based assay to evaluate protease inhibitors. While multiple assays 55
exist to evaluate protease inhibitors, an assay of the nature we envisioned has clear advantages, as 56
it requires minimal upfront cost or effort, is accessible to many biomedical research labs, does not 57
involve the use of live virus, and requires no specialized reporter to read out protease activity. In 58
contrast, in vitro protease assays using purified protein have formed the backbone of inhibitor 59
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
4
screening, but require upfront efforts to isolate the pure protease and are not conducted under 60
physiologic cellular conditions13,14. In addition, if one desires to identify broad-acting coronavirus 61
inhibitors, one must purify and identify experimental conditions suitable for testing each protease in 62
vitro. An alternative approach for identifying protease inhibitors is the use of live virus which is 63
performed under more biologically relevant conditions, assuming relevant host cell systems can be 64
identified, but requires intensive safety training and specialized biosafety protocols15. For many 65
coronaviruses, no live virus assay exists, limiting the ability to test compounds within mammalian cell 66
systems to a small subset of all coronaviruses16. Furthermore, compounds with activity against live 67
virus may function through a number of mechanisms other than protease inhibition which cannot be 68
readily determined, and may lead to undesired off-target activities which are not realized until much 69
later in the drug development process17,18. 70
71
Here, we report a mammalian cell-based assay to identify coronavirus 3CLpro inhibitors that does 72
not require the use of live virus. We demonstrate the utility of the assay using the SARS-CoV-2 73
3CLpro, with EC50 values obtained from the assay showing good concordance with traditional live 74
virus testing for multiple compounds. We next establish the generality of the approach by testing a 75
diverse set of 3CLpro enzymes. Finally, we perform a small molecule screen, along with structure-76
activity profiling of a set of compounds to find those with enhanced antiviral activity. The presented 77
data support the use of our assay system for the discovery of small molecule 3CLpro inhibitors and 78
the rapid characterization of their activity across multiple coronaviruses to identify those with broad 79
inhibitory activity. 80
81
82
Results 83
Expression of the SARS-CoV-2 3CLpro in HEK293T cells results in cell toxicity and can be 84
used as a readout of drug activity 85
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
5
Prior work has shown that the exogenous expression of viral proteases within cellular systems can 86
result in growth suppression or cell death. Motivated by this work, we sought to determine the effect 87
of expressing the SARS-CoV-2 3CLpro in HEK293T cells. Utilizing a cost-effective crystal-violet-88
based approach to quantify cell abundance, we observed that expression of SARS-CoV-2 3CLpro 89
results in significant growth inhibition as compared to the expression of a control protein, enhanced 90
yellow fluorescent protein (EYFP) (Fig. 1a-b)19. This suppression of growth was dependent upon the 91
catalytic function of the enzyme, as mutating cysteine 145, which is essential for the enzyme’s 92
peptidase activity, abolished the growth defect (Fig. 1a-b). We sought to determine if the observed 93
growth defect could be rescued by incubating cells with GC376, a feline coronavirus inhibitor that 94
was recently reported to have activity against SARS-CoV-2 3CLpro20. In comparison to untreated 95
control cells, the addition of GC376 led to a robust increase in cell growth (Fig. 1c-d). 96
97
Compound rescue of transfected 3CLpro cytotoxicity is similar to results obtained with live 98
virus 99
We next tested if this transfection-based assay could be used to determine compound EC50 values 100
and whether the values showed any correlation with those obtained with live virus. After incubating 101
SARS-CoV-2 3CLpro transfected cells with a range of GC376 concentrations, we calculated an EC50 102
of 3.30 µM, which is similar to published values reported using live virus on Vero E6 (EC50 2.2 µM, 103
0.9 µM, 0.18 µM, 4.48 µM) and Vero 76 cells (EC50 3.37 µM) (Fig. 2a and Table 1)20–24. We 104
investigated the assay’s tolerance to deviation by varying the amount of plasmid transfected or the 105
number of cells seeded into wells containing compound (Supplementary Fig. 1). In all cases, the 106
assay was robust to variation, delivering a similar EC50 for GC376 across all conditions. We also 107
tested an orthogonal method of quantifying cell abundance based on fluorescence microscopy and 108
observed agreement with the results obtained with crystal violet staining (Supplementary Fig. 2). 109
This suggests that the assay system is robust to variation and provides consistent results. 110
111
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
6
We next conducted dose-response profiling for two additional SARS-CoV-2 3CLpro inhibitors, 112
compound 4 and compound 11a, and observed reversal of the toxic effect of the protease in a dose-113
dependent manner (Fig. 2b-c)25,26. In agreement with the results obtained with GC376, the EC50 114
value for compound 4 was comparable to those obtained with live virus, 0.98 µM and 3.023 µM, 115
respectively (Table 1)24. Unexpectedly, we calculated an EC50 of 6.89 µM for 11a, which is 116
approximately 10-fold higher than the literature reported value of 0.53 µM, based on viral plaque 117
assay26. We have noticed that literature reported EC50 values from live virus testing could range over 118
an order of magnitude depending on the exact method employed, as is the case for GC376 (Table 119
1). To resolve this discrepancy between the transfection-based approach and the live virus assay, 120
we conducted live virus testing of 11a using the commonly employed readout of cytopathic effect in 121
Vero E6 cells and observed closer concordance with our transfection-based results (Supplementary 122
Figure 3 and Table 1)20,24,27. To measure the toxicity of each compound, we exposed EYFP-123
transfected cells to each molecule and determined CC50 values (Fig. 2). We also calculated the 124
selectivity index (SI) for each compound tested in this study (Supplementary Table 1). 125
126
We hypothesized that the assay would be able to distinguish between compounds that are only 127
active on the purified SARS-CoV-2 3CLpro and those that are able to inhibit the live virus through 128
protease inhibition. In general, we observe concordance between compounds showing activity within 129
this transfection-based 3CL assay and live virus studies (Supplementary Fig. 4a-e)13. However, 130
within the assay, we did not observe activity for ebselen, a small molecule with demonstrated activity 131
against the SARS-CoV-2 3CLpro in vitro and activity against the SARS-CoV-2 live virus 132
(Supplementary Fig. 4f). We suggest that this may be due to ebselen targeting more than 3CLpro 133
within the live virus assay, which is in line with work showing that ebselen is highly reactive and 134
readily forms selenosulfide bonds with numerous proteins including the SARS-CoV-2 papain-like 135
protease (PLP)18,28,29. 136
137
Demonstrating assay compatibility across a range of coronavirus 3CLpro enzymes 138
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
7
We hypothesized that this assay may be used to study other coronavirus 3CLpros to enable users to 139
identify broad-acting inhibitors, as constructs containing other 3CLpro enzymes could be readily 140
synthesized. To test the assay’s generality, we created expression constructs for 3CL proteases 141
from five other coronaviruses (SARS-CoV, MERS-CoV, Bat-CoV-HKU9, HCoV-NL63 and IBV) with 142
variable amino acid identity compared with SARS-CoV-2 3CLpro (Supplementary Fig. 5a). For each 143
of these proteases, we confirmed that expression in mammalian cells resulted in toxicity that is 144
dependent upon the enzyme’s catalytic activity (Supplementary Fig. 5b). Next, we tested GC376, 145
compound 4, and 11a across this panel of proteases. GC376, a drug originally identified for use 146
against the Feline Infectious Peritonitis virus, showed EC50 <10 µM for the most, but not all of 147
proteases tested30. Unexpectedly, compound 4, which was originally designed as a SARS-CoV 148
3CLpro inhibitor showed particular potency against IBV 3CLpro (EC50 = 0.058 µM) along with broad 149
activity (EC50 <10 µM) for all other 3CL proteases tested. In contrast to GC376 and compound 4, 11a 150
had a relatively narrow activity spectrum with EC50 <10 µM against only SARS-CoV and SARS-CoV-151
2 3CLpro enzymes (Fig. 3). Of note, in all cases where previous live virus data was available, the 152
EC50 values obtained from this transfection-based assay were similar (Table 1). 153
154
Rapid testing to identify SARS-CoV-2-3CL protease inhibitors 155
Having further determined the assay’s ability to examine the effects of active individual compounds, 156
we sought to determine its suitability for small molecule screening. Before performing the screen, we 157
optimized the testing parameters to ensure suitable performance characteristics (Supplementary Fig. 158
6 and Methods)31. We compiled a collection of 162 diverse protease inhibitors, along with 159
compounds with reported in vitro activity against 3CLpro enzymes or structural similarity to known 160
3CLpro inhibitors (Supplementary Table 2 and Supplementary Table 3). Of the nearly 200 161
compounds tested, two potent hits were identified, GC373 and GRL-0496 (Fig. 4a, Supplementary 162
Table 2)32. 163
164
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
8
We noted that GC373 is structurally similar to its prodrug GC376, except for the change of the 165
bisulfide salt adduct to an aldehyde warhead22,33. Additional testing of GC373 revealed it to have a 166
similar EC50 as GC376 in both the transfection assay and when tested against live SARS-CoV-2 167
virus, suggesting that the differences in structure has a minimal effect on their potency 168
(Supplementary Fig. 7 and Table 1), although solubility may be affected33. The other hit from the 169
screen, GRL-0496, shares structural similarity to several other compounds within the library, one of 170
which is a previously reported 3CLpro inhibitor (MAC-5576) that failed to show activity against 171
SARS-CoV-2 in a live virus assay24,34. Additional testing of GRL-0496 revealed it to have an EC50 of 172
5.05 µM against SARS-CoV-2 3CLpro within our transfection-based assay (Fig. 4c). To verify GRL-173
0496’s activity, we tested it against live SARS-CoV-2 virus, and confirmed its potency (EC50 = 9.12 174
µM) (Fig. 4d). We next tested GRL-0496 against the full panel of 3CLpro enzymes we had previously 175
examined and observed a narrow range of activity, with EC50 <10 µM only observed against SARS-176
CoV 3CLpro and SARS-CoV-2 3CLpro, in agreement with previous live virus testing (Supplementary 177
Fig. 8 and Table 1)27. We also tested GC373 against the full panel of 3CLpro enzymes and observed 178
concordance with GC376, with SARS-CoV 3CLpro, MERS-CoV 3CL pro, and IBV 3CL pro 179
demonstrating an EC50 <10 µM (Supplementary Fig. 8). 180
181
Discussion 182
Given the potential for protease inhibitors in the treatment of viral illnesses, small molecule inhibitors 183
of coronavirus 3CL proteases represent a promising avenue for treating infections caused by this 184
large family of viruses. Here, we present a simplified assay to identify candidate inhibitors under 185
physiologic cellular conditions. This approach presents significant advantages over other methods to 186
detect 3CL protease inhibitory activity with its ease of use and ability to be performed with equipment 187
and reagents commonly available to many biomedical research laboratories. While conventional 188
methods for identifying 3CL protease inhibitors make use of in vitro purified protease, the isolation of 189
sufficiently pure enzyme in its native state can be costly and labor intensive. Furthermore, assays 190
using purified protease fail to consider cell permeability and the influence of the extracellular and 191
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
9
intracellular milieu on compound activity. In comparison to live virus-based assays, the outlined 192
approach does not require extensive biosafety containment. These data also suggest that the 193
approach described here is applicable to a number of coronaviruses for which live virus assays may 194
not be available or would be deemed ethically challenging to be performed even with extensive 195
biosafety infrastructure35,36. Finally, because the phenotype assayed within this approach is driven 196
solely by protease activity, it may enable the distinction between compounds with multiple biological 197
targets and subsequent potential for off-target toxicity from those that function primarily as 3CLpro 198
inhibitors. 199
200
A number of surrogate assays that aim to identify molecules with activity against proteins encoded 201
by SARS-CoV-2 have recently been reported37–39. These assays were developed to provide 202
physiologically meaningful molecule testing without needing to use live SARS-CoV-2 cultures in 203
order to allow for rapid testing and widespread adoption to labs without necessary safety 204
infrastructure. The aforementioned assays have focused primarily on identifying neutralizing 205
antibodies targeted at the SARS-CoV-2 spike (S) protein. To our knowledge, a reporter-free 206
surrogate assay to identify coronavirus 3CL protease inhibitors validated with multiple known 207
coronavirus 3CL protease inhibitors has not been reported. 208
209
Within the literature, EC50 values obtained for a 3CLpro inhibitor against live virus can show a broad 210
range of potency, with some compounds demonstrating EC50 values across multiple orders of 211
magnitude (Table 1). These differences appear to be driven by variation in experimental setup such 212
as cell line used, assay readout, incubation period, and initial concentration of virus added. While we 213
have observed agreement between the EC50 values obtained from the described transfection-based 214
method and those reported in the literature, given the differences in EC50 across assays, we suggest 215
caution when comparing results across studies. By developing this transfection-based 3CLpro 216
testing platform, we hope to facilitate the discovery of new coronavirus inhibitors while also 217
facilitating the comparison of existing inhibitors within a single simplified assay system. Furthermore, 218
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
10
we propose that this cellular protease assay system could be industrialized to screen and optimize a 219
large number of compounds to discover potential treatments for future viral pandemics. 220
221
Methods 222
Cell Lines and Cell Culture 223
HEK293T and HEK293 cells used in this study were obtained from ATCC. Cells were maintained at 224
37°C in a humidified atmosphere with 5% CO2. HEK293T and HEK293 cells were grown in 225
Dulbecco’s Modified Eagle Medium (DMEM, Invitrogen) which was supplemented with 10% fetal 226
bovine serum (Gibco) and penicillin-streptomycin (Invitrogen). HEK293T and HEK293 cells were 227
confirmed to be free of mycoplasma contamination with the Agilent MycoSensor PCR Assay Kit. 228
To obtain HEK293 cells stably expressing EYFP, cells were co-transfected with EYFP plasmids 229
harboring the piggyBac transposon (pPB bsr2-EYFP) (Yusa et al., 2009) and pCMV-mPBase 230
(mammalian codon-optimized PBase) encoding a piggyBac transposase using Lipofectamine 2000 231
(Invitrogen) according to the manufacturer’s instructions. One day after transfection, the transfected 232
cells were selected with 10 µg/mL of blasticidin (Invitrogen). 233
234
Transfections and Drug Selections 235
24 h prior to transfection, 293T cells were seeded at 65-75% confluency into 24-well plates coated 236
for 30 min with a 1 mg/mL solution of poly-D-lysine (MP Biomedicals Inc.) and washed with PBS 237
(Gibco) once prior to media addition. The next day, 500 ng of 3CLpro expression plasmid, unless 238
otherwise stated, was incubated with Opti-MEM and Lipofectamine 2000 for 30 min at room 239
temperature prior to dribbling on cells, as per manufacturer’s protocol. For plating into drug 240
conditions, 20 h after transfection, cells were washed once with PBS and 200 µL Trypsin-EDTA 241
0.25% (Gibco) was added to cells to release them from the plate. Trypsinized cell slurry was pipetted 242
up and down repeatedly to ensure a single cell suspension. 96-well plates were coated with poly-D-243
lysine, either coated manually with 1 µg/mL poly-D-lysine in PBS solution for 30 min or purchased 244
pre-coated with poly-D-lysine (Corning). Wells were filled with 100 µL of media ± drug and 1 µg/mL 245
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
11
puromycin and were seeded with 9 µL of trypsinized cell slurry. For higher throughput experiments, 246
multiple individually transfected wells of a 24-well plate were combined after trypsinization and prior 247
to seeding in drug. After seeding into wells containing drug and puromycin, cells were incubated for 248
48 h unless otherwise specified. 249
250
Plasmids 251
All vectors used in this study were cloned into the pLEX307 backbone (Addgene #41392) using 252
Gateway LR II Clonase Enzyme mix (Invitrogen). 3CL proteases used in this study were generated 253
using gene fragments ordered from Twist Biosciences. Inactive 3CL proteases were generated by 254
site directed mutagenesis of the essential catalytic cysteine. DNA was transformed into NEB 10-beta 255
high efficiency competent cells. Sanger sequencing to verify proper inserts were done for all 256
plasmids used in this study (Genewiz). 257
258
Plasmid DNA was isolated using standard miniprep buffers (Omega Biotek) and silica membrane 259
columns (Biobasic). To reduce batch-to-batch variability between plasmid DNA isolations and its 260
subsequent impact on transfection efficiency, multiple plasmid DNA extractions were conducted in 261
parallel, diluted to 50 ng/µL and pooled together. 262
263
Crystal Violet Staining and Quantification 264
The crystal violet staining protocol was adapted from Feoktistova et. al.19 Briefly, after compound 265
incubation with 3CLpro expressing cells in 96-well plates, the medium was discarded and cells were 266
washed once with PBS. Cells were incubated with 50 µL of crystal violet staining solution (0.5% 267
crystal violet in 80% water and 20% methanol) and rocked gently for 30 min. The staining solution 268
was removed and cells were washed four times with water using a multichannel pipette. Stained 269
cells were left to dry for ≥4 h on the laboratory bench or within the chemical hood. The crystal violet 270
staining solution was eluted by the addition of 200 µL of methanol over the course of 30 min with 271
gentle rocking. Plates were sealed with parafilm to mitigate methanol evaporation. 100 µL of eluted 272
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
12
stain from each well was transferred to a new 96-well plate for reading in a Tecan Infinite F50 plate 273
reader. Absorbance was measured at 595 nm twice and values were averaged between replicate 274
measurements. Blank wells were included in each batch of experiments, and absorbance values 275
were normalized by background levels of staining from blank wells. 276
277
Statistical Analysis of Dose Response Curves 278
For analysis of crystal violet staining experiments, relative growth was calculated from background 279
normalized absorbance values. Test wells containing drug were divided by average background 280
normalized values from wells where cells were expressing protease and exposed to vehicle, when 281
available. Otherwise, values were normalized by values from protease-expressing cells exposed to 282
the lowest concentration of drug included in the dose-response curve. When there were significant 283
deviations from protease-expressing cells exposed to no drug and protease-expressing cells 284
exposed to lowest concentrations of drug included in the dose-response curve, experiments were 285
repeated with normalization by protease-expressing cells exposed to no drug. CC50 values were 286
calculated in Prism using the nonlinear regression functionality and derived from dose-response 287
curves with EYFP transfected cells. A nonlinear curve fitting function accounting for variable curve 288
slopes was calculated by plotting the normalized response as a function of log(compound). Similarly, 289
EC50 values were calculated in GraphPad Prism also using the nonlinear regression functionality. A 290
nonlinear curve fitting function measuring the stimulatory response of a compound as a function of 291
an unnormalized response was used to calculate the EC50. All reported values were confirmed to not 292
have ambiguous curve fitting. The 95% confidence interval of EC50 calculations was also calculated 293
and included. 294
295
For analysis of live virus experiments, EC50 values were determined by fitting a nonlinear curve to 296
the data with the assumption of a normalized response (GraphPad Prism). Cells were confirmed as 297
mycoplasma negative prior to use. 298
299
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
13
Compound Screening 300
For screening condition optimization, we measured the Z-Factor for replicates of positive controls 301
GC376, tested at 50 µM, and compound 4, tested at 20 µM. Replicate measurements were recorded 302
for DMSO negative controls and positive control compounds after 48, 72, and 96 h of incubation with 303
drug. Background normalized crystal violet absorbance values at each timepoint were collected. 304
305
During the drug screen, within each of the four plates screened, two positive controls wells were 306
included to ensure assay reliability, along with several wells with the negative control 0.1% DMSO 307
condition. All compounds were screened at 10 µM resuspended in DMSO (Fisher Scientific). 308
For hit selection, we employed a robust z-score method. We first normalized data using a robust z-309
score that uses median and median absolute deviation (MAD) instead of mean and standard 310
deviation. We then used a threshold of 3.5 MAD to determine which drugs rescued the cytotoxicity 311
imposed by expression of the viral protease40. 312
313
Live Virus Assay 314
The SARS-CoV-2 strain 2019-nCoV/USA_WA1/2020 was grown and titered in Vero-E6 cells. One 315
day before the experiment, Vero-E6 cells were seeded at 30,000 cells/well in 96 well-plates. Serial 316
dilutions of the test compound were prepared in media (EMEM + 10% FCS + 317
penicillin/streptomycin), pipetted onto cells, and virus was subsequently added to each well at an 318
MOI of 0.2. Cells were incubated at in a humidified environment at 37 °C with 5% CO2 for 72 h after 319
addition of virus. Cytopathic effect was scored by independent, blinded researchers. The reported 320
cytopathic effect value represents the average from two independent reviewers. Percent Inhibition 321
was calculated by comparison to control wells with no inhibitor added. All live virus experiments were 322
conducted in a biosafety level 3 lab. 323
324
Microscopy 325
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
14
Cells plated on a 96-well plate (Greiner Bio-OneTM) and washed once immediately before imaging. 326
EYFP fluorescence imaging was performed using an Axio Observer 7 microscope (Zeiss) equipped 327
with a Plan-Apochromat 10X objective (0.45 N.A.) with 1-by-1 pixel binning. Optical Illumination bias 328
was empirically derived by sampling background areas and subsequently used to flatten images. 329
After a global background subtraction, cell density was calculated based on area of EYFP intensity. 330
331
Compounds and Chemical Synthesis 332
GC376 was purchased from Aobious. Myrecetin, rupintrivir, grazoprevir, saquinavir, fosamprenavir, 333
indinavir, apigenin, quercetin, famotidine, MDL28170, bicailein, betrixaban, and amentoflavone were 334
purchased from Fisher Scientific. Tipranavir was purchased from Cayman Chemical. MAC5576, 335
MAC22272, MAC8120, MAC30731, BTB07404, BTB07408, MWP00332, BTB07417, MWP00508, 336
MWP00333, BTB07407, SPB08384, SPB06613, SPB06636, SPB06591, SPB06593, MWP00709, 337
CC42746, BTB07789, BTB07420, MWP00710, BTB07421, SCR00533, and SEW03089 were 338
purchased from Maybridge. GRL-0496 and GRL-0617 were purchased from Focus Biomolecules. 339
AZVIII-40A (1,2-Benzisothiazol-3(2H)-one) was purchased from Alfa Aesar. Other protease inhibitors 340
in table below were purchased from TargetMol. 341
342
CAS Number Drug Name CAS Number Drug Name 1226781-44-7 Omarigliptin 134381-21-8 Epoxomicin 79183-19-0 Apoptosis Activator 2 110044-82-1 MG101 3731-52-0 Picolamine 125697-93-0 lavendustin C 541-91-3 Muscone 729607-74-3 BMS707035 60-23-1 2-Aminoethanethiol 630420-16-5 Asunaprevir 51146-56-6 Dexibuprofen 76684-89-4 Loxistatin Acid 3416-24-8 Glucosamine 1025015-40-0 GK921 56974-61-9 Gabexate mesylate 292632-98-5 L-685,458 7481-89-2 Zalcitabine 202138-50-9 Tenofovir Disoproxil Fumarate 2016-88-8 Amiloride hydrochloride 937174-76-0 GSK690693 945667-22-1 Saxagliptin hydrate 1256388-51-8 Ledipasvir 668270-12-0 Linagliptin 960374-59-8 ONX0914 486460-32-6 Sitagliptin 1401223-22-0 PI1840 136-77-6 Hexylresorcinol 13552-72-2 (+)-Isocorydine hydrochloride
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
15
497-76-7 Arbutin 697797-51-6 UAMC 00039 dihydrochloride 908-54-3 Diminazene Aceturate 1402727-29-0 PE859
6419-36-9 3-Pyridylacetic acid hydrochloride 847925-91-1 RO4929097
81110-73-8 Racecadotril 254750-02-2 Emricasan 50924-49-7 Mizoribine 161314-82-5 CGS 27023A 7414-83-7 Sodium etidronate 150080-09-4 Talabostat mesylate 103-16-2 Monobenzone 1441674-54-9 Ledipasvir acetone 1180-71-8 Limonin 130370-60-4 Batimastat 472-15-1 Betulinic acid 54857-86-2 TOFA 329-98-6 PMSF 1009734-33-1 HZ1157 42017-89-0 Fenofibric acid 188062-50-2 Abacavir sulfate 196597-26-9 Ramelteon 127373-66-4 Sivelestat 155213-67-5 Ritonavir 1132935-63-7 Dasabuvir 850649-62-6 Alogliptin Benzoate 20575-57-9 Calycosin 179324-69-7 Bortezomib 673-22-3 4-Methoxysalicylaldehyde 546-88-3 Acetohydroxamic acid 111-20-6 Sebacic acid 129618-40-2 Nevirapine 53936-56-4 Deoxyarbutin 192725-17-0 Lopinavir 490-78-8 2-5-dihydroxyacetophenone 39809-25-1 Penciclovir 29700-22-9 Oxyresveratrol 916170-19-9 AOB2796 88321-09-9 Aloxistatin 176161-24-3 Maribavir 864953-29-7 Fostemsavir 1029877-94-8 Trelagliptin succinate 519055-62-0 Tasisulam 1201902-80-8 MLN9708 425386-60-3 Semagacestat 354812-17-2 SC514 35943-35-2 Triciribine 1072833-77-2 Ixazomib 331862-41-0 IMR-1A 871038-72-1 Raltegravir potassium 310456-65-6 IMR1 863329-66-2 PSI6206 210344-98-2 Z-IETD-FMK 82009-34-5 Cilastatin 1624602-30-7 VR23 480-18-2 Taxifolin 161814-49-9 Amprenavir 82956-11-4 Nafamostat mesylate 1312782-34-5 AA26-9 1009119-65-6 Daclatasvir dihydrochloride 1051375-16-6 Dolutegravir 635728-49-3 Darunavir Ethanolate 1026785-59-0 Lomibuvir 142880-36-2 Ilomastat 112246-15-8 Ginsenoside Rh2 697761-98-1 Elvitegravir 256477-09-5 UK371804 1051375-19-9 Dolutegravir sodium 147859-80-1 CA-074 methyl ester 84687-43-4 Astragaloside IV 1404437-62-2 ML281 7770-78-7 Arctigenin 591778-68-6 CP 640186 83-48-7 Stigmasterol 5631-68-5 Hydroumbellic acid 478-01-3 Nobiletin 831-61-8 Ethyl gallate 34157-83-0 Celastrol 2469-34-3 Senegenin 29031-19-4 Glucosamine sulfate 28831-65-4 lithospermic acid
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
16
27409-30-9 Picroside I 209984-56-5 Dibenzazepine 848141-11-7 Alvelestat 209984-57-6 LY411575 128-53-0 N-Ethylmaleimide 1221573-85-8 Paritaprevir 208255-80-5 DAPT 1190307-88-0 Sofosbuvir 865759-25-7 Trelagliptin 1421438-81-4 Crenigacestat 187389-52-2 Z-VAD(OMe)-FMK 1338225-97-0 Doravirine 514-10-3 Abietic Acid 847499-27-8 Delanzomib 229975-97-7 Atazanavir sulfate 25406-64-8 Morroniside
136470-78-5 Abacavir 20633-67-4 Calycosin-7-O-beta-D-glucoside
354813-19-7 Balicatib 59870-68-7 Glabridin 868540-17-4 Carfilzomib 58749-22-7 Licochalcone A 198904-31-3 Atazanavir 1377049-84-7 Velpatasvir 274901-16-5 Vildagliptin 402957-28-2 Telaprevir 244767-67-7 Dapivirine 603139-19-1 Odanacatib 292605-14-2 SB-3CT 206361-99-1 Darunavir 179461-52-0 PD 151746 850876-88-9 Danoprevir 315183-21-2 PAC1 159989-65-8 Nelfinavir Mesylate 59721-29-8 Camostat mesilate 935888-69-0 Oprozomib 154598-52-4 Efavirenz 30827-99-7 AEBSF hydrochloride
1192224-24-0 Des(benzylpyridyl) Atazanavi 273404-37-8 Belnacasan
1194044-20-6 LY2811376 210344-95-9 Z-DEVD-FMK 313967-18-9 FLI06 197855-65-5 Z-FA-FMK 218156-96-8 SRPIN340 149488-17-5 Trovirdine 7497-07-6 NSC 405020 133407-82-6 MG132 103476-89-7 Leupeptin Hemisulfate 1051375-10-0 Cabotegravir 57-11-4 Stearic acid 1146699-66-2 Avagacestat 343
All other compounds used in the study were synthesized and quality checked according to the 344
following protocols. 345
346
347
2-(methylthio)nicotinonitrile (xx-1) The title compound was prepared according to a published 348
procedure; spectral data are in agreement with literature values41. 349
N
CN
SCH3
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
17
1H NMR (500 MHz, CDCl3) δ 8.60 (dd, J = 5.0, 1.8 Hz, 1H), 7.79 (dd, J = 7.7, 1.8 Hz, 1H), 7.07 350
(dd, J = 7.7, 4.9 Hz, 1H), 2.64 (s, 3H). 351
13C NMR (126 MHz, CDCl3) δ 163.7, 152.2, 140.6, 118.4, 115.7, 107.5, 13.4. 352
HRMS High accuracy (ASAP): Calculated for C7H6N2S (M+H)+: 151.0330; found: 151.0327. 353
354
355
2-(methylsulfonyl)nicotinonitrile (xx-2) The title compound was prepared from xx-1 according to 356
a modified procedure from the literature42. 357
358
Pyridine xx-1 (87 mg, 0.58 mmol, 1 equiv) was weighed into 50 mL round bottom flask equipped 359
with a stir bar. The solid was dissolved in 10 mL of anhydrous MeOH, followed by portion-wise 360
(usually 3 portions) additions of mCPBA (500 mg, 2.9 mmol, 5 equiv). Substrate conversion was 361
monitored via TLC analysis (70% EtOAc:Hex to 100% EtOAc), with up to an additional 5 equiv 362
of mCPBA added if necessary. Upon complete conversion of starting material, the reaction was 363
quenched with 20 mL of saturated aqueous NaHCO3, diluted with an additional 20 mL of DCM. 364
The layers were separated and the organic layer was further washed (2x) with 10 mL of of 365
saturated aqueous NaHCO3. The organic layer was then dried in vacuo and purified via silica 366
gel column chromatography (100% EtOAc to 5% MeOH:EtOAc) to yield 51 mg (48% yield) of 367
the desired sulfone as a white solid. 368
1H NMR (400 MHz, CDCl3) δ 8.86 (dd, J = 4.8, 1.6 Hz, 1H), 8.26 (dd, J = 7.9, 1.6 Hz, 1H), 7.70 369
(dd, J = 7.9, 4.8 Hz, 1H), 3.39 (s, 3H). 370
13C NMR (101 MHz, CDCl3) δ 159.8, 151.8, 143.8, 126.8, 113.2, 107.4, 40.1. 371
HRMS High accuracy (ASAP): Calculated for C7H6N2O2S (M+H)+: 183.0228; found: 183.0223. 372
N
CN
S CH3O O
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
18
373
374
2-(methylthio)-6-(trifluoromethyl)nicotinonitrile (xx-3) The title compound was prepared following 375
the published procedure, the product (yellow solid) was carried towards the synthesis of xx-4 376
without further purification41. 377
378
379
2-(methylsulfonyl)-6-(trifluoromethyl)nicotinonitrile (xx-4) The title compound was prepared from 380
xx-3 according to a modified procedure from the literature42. 381
382
Pyridine xx-2 (105 mg, 0.50 mmol, 1 equiv) was weighed into 50 mL round bottom flask 383
equipped with a stir bar. The solid was dissolved in 10 mL of anhydrous MeOH, followed by 384
portion-wise (usually 3 portions) additions of mCPBA (0.86 mg, 5 mmol, 10 equiv). Substrate 385
conversion was monitored via TLC analysis (40% EtOAc:Hex to 60% EtOAc). Upon complete 386
conversion of starting material, the reaction was quenched with 20 mL of saturated aqueous 387
NaHCO3, diluted with an additional 20 mL of DCM. The layers were separated and the organic 388
layer was further washed (2x) with 10 mL of of saturated aqueous NaHCO3. The organic layer 389
was then dried in vacuo and purified via silica gel column chromatography (40% to 80% 390
EtOAc:Hex) to yield 38 mg (30% yield) of the desired sulfone as a white solid. 391
1H NMR (400 MHz, CDCl3) δ 8.49 (dd, J = 8.1, 0.7 Hz, 1H), 8.06 (d, J = 8.1 Hz, 1H). 392
13C NMR (101 MHz, CDCl3) δ 160.4, 149.7 (q, J = 37.6 Hz), 146.0, 123.7 (q, J = 2.3 Hz), 120.0 393
(q, J = 275.5 Hz), 112.1, 109.7, 39.7. 394
N
CN
SCH3CF3
N
CN
S CH3O O
CF3
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
19
19F NMR (376 MHz, CDCl3) δ -68.28. 395
HRMS High accuracy (ASAP): Calculated for C8H5F3N2O2S (M+H)+: 251.0102; found: 251.0104. 396
397
Compound 11a was synthesized according to the specified protocol in Dai. et. al. 202026. Compound 398
11a was confirmed by LCMS with m/z = 453 (M+1) and 451 (M-1). 399
400
Compound 4 was synthesized according to the procedure described in Yang et. al 200625. AZVIII-401
34D was formed as a byproduct (15%) in the synthesis of Compound 4 and was isolated by RP 402
HPLC. 1H NMR (400 MHz, CDCl3) 7.88 (s, 1H), 7.44-7.34 (m, 5H), 6.80 (d, 1H, J = 15.4 Hz), 6.69 (s, 403
1H), 6.16-5.87 (m, 2H), 5.12 (s, 2H), 4.76 (s, 1H), 4.42 (s, 1H), 4.31-3.99 (m, 4H), 3.39 (s, 2H), 2.40-404
1.45 (m, 8H), 1.29–1.20 (m, 12H), 1.13-1.02 (m, 3H), 1.00 – 0.89 (m, 6H). MS M+H = 631. 405
406
AZVIII-38, AZVIII-30, and AZVIII-42 were synthesized according the procedures described in Mou 407
et. al. 200843. AZVIII-37A was synthesized as described in Prior et al. 2013. AZVIII-33B was 408
synthesized according to the method described in Amblard et. al. 201844. 409
410
AZVIII-41A was synthesized according to the procedure described for synthesizing AZVIII-33B in 411
Amblard et. al. 2018 and substituting Z-Leu-OH for BOC-Leu-OH44. 1H NMR (400 MHz, CDCl3) 7.77 412
(s, 1H), 7.41-7.27 (m, 5H), 6.80 (dd, 1H, J = 5.5, 15. 6 Hz), 5.98-5.84 (m, 2H), 5.15-5.03 (m, 2H), 413
4.55 (bd s, 1H), 4.27 (bd s, 1H), 4.23-4.07 (m, 3H), 3.38-3.23 (m, 2H), 2.56–1.40 (m, 8H), 1.33 – 414
1.20 (m, 3H), 0.95 (s, 6H). MS M+H = 474. 415
416
AZVII-43A was synthesized by treatment of a dichloromethane solution of AZVIII-30 and AZVIII-40A 417
with triphenyl phosphine and diethyl azodicarboxylate. 1H NMR (400 MHz, CDCl3) 7.91 (d, 1H, J = 418
8.0), 7.76 (d, 1H, J = 8.0), 7.55-7.34 (m, 2H), 6.16 (bd s, 1H), 5.29 (bd s, 1H), 4.73-4.39 (m, 2H), 419
4.35-3.96 (m, 1H), 3.58-3.16 (m, 2H), 3.79-3.43 (m, 2H), 3.39-3.91 (m, 1H), 1.98-1.77 (m, 1H), 1.73-420
1.57 (m, 1H), 1.42 (s, 9H). MS M+H = 392. 421
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
20
422
423
AZVIII-44B, AZVIII-44D, AZVIII-44E, AZVIII-44H, AZVIII-49C, and AZVIII-49F were synthesized by 424
treatment of the appropriate arylalkylamine or heteroarylalkylamine with the corresponding 425
arylsulfonyl chloride or heteroarylsulfonyl chloride and diisopropylethylamine in dichloromethane. 426
AZVIII-57D was synthesized by treatment of AZVIII-44D with methyl bromoacetate and potassium 427
carbonate in dimethylformamide to give the corresponding ester which was reduced to the 428
corresponding alcohol by treatment with lithium borohydride in tetrahydrofuran. The alcohol was 429
then treated with 1,1,1-tris(acetyloxy)-1,1-dihydro-1,2-benziodoxol-3-(1H)-one and sodium 430
bicarbonate in dichloromethane to give to the corresponding aldehyde AZVIII-57D. AZVIII-57G was 431
synthesized from N-benzyl-4-methoxybenzenesulfonamide according to the procedure used to 432
synthesize AZVIII-57D. 433
434
GC373 was synthesized from GC376 by converting the bisulfite group to an aldehyde group by 435
treatment with aqueous sodium bicarbonate. To a solution of 5.31 mg of GC376 in 200 µL H2O, 2 µL 436
O
O
NH
NHO
NS OO
O
NH
NHO
OH
AZVIII-30
HNS
O+
AZVIII-40A
DEAD
P(Ph)3
AZVIII-43A
CH2Cl2
R NH2 ’R SO2Cl+
R NH
S R’R, R’ = aryl, heteroaryl
N
CH2Cl2 AZVIII-44B AZVIII-44D AZVIII-44E AZVIII-44H AZVIII-49C AZVIII-49F
Br CO2Me
K2CO3
DMF
N S
X
O
O
X = H, OMe
N S
X
OH
LiBH4
THF
N S
X
O
DMP
NaHCO3
CH2Cl2 AZVIII-57D X = HAZVIII-57G X = OMe
O
O
O
O
O
O
O
O
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
21
of saturated NaHCO3 was added. The cloudy mixture was extracted with CH2Cl2, dried over Na2SO4, 437
and concentrated in vacuo to give GC373 as a colorless oil (4.0 mg). 438
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
22
Data availability 439
All reagents generated in this study are without restriction. Plasmids generated in this study will be 440
deposited to Addgene. Source data for all figures are provided with this manuscript online. All 441
statistics were performed using Prism v.8.4.2. 442
443
Acknowledgements 444
Compound 11a was synthesized and generously provided by Shi-Xian Deng, Columbia University. 445
This work was supported by a grant from the Jack Ma Foundation to D.D.H. and A.C. and by grants 446
from Columbia Technology Ventures and the Columbia Translational Therapeutics (TRx) program to 447
B.R.S. Further support for these studies comes from a pilot grant award from the Herbert Irving 448
Comprehensive Cancer Center in partnership with the Irving Institute for Clinical and Translational 449
Research at Columbia University to H.Y. and A.C. A.C. is also supported by a Career Awards for 450
Medical Scientists from the Burroughs Wellcome Fund. S.J.R is supported by NIH grant 451
F31NS111851. S.I. is supported by NIH grant T32AI106711. 452
453
Author contributions 454
S.J.R., S.I, and A.C conceived the project. S.J.R., S.I., B.R.S., D.D.H., and A.C. planned and 455
designed the experiments. S.J.R. performed crystal violet-based assays. S.J.R. and S.K. performed 456
the HEK293-EYFP based assays. S.J.R., S.I., and S.J.H. cloned plasmids. A.Z. and H.L. 457
synthesized compounds and provided compound structure information for synthesized compounds. 458
N.E.S.T. and T.R. synthesized compounds NT-1-21, NT-1-24, and NT-1-32. M.S.N. and Y.H. 459
conducted the live virus assays. S.K. and H.Y. performed imaging and created the HEK293-EYFP 460
cell line. S.J.R. and S.M. conducted data analysis. S.J.R, S.I., S.J.H, and A.C. wrote the manuscript 461
with input from all authors. 462
463
Competing interests 464
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
23
S.I., H.L., A.Z., B.R.S., A.C., and D.D.H. are inventors on a patent application submitted based on 465
some of the molecules described in this work. B.R.S. is an inventor on additional patents and patent 466
applications related to small molecule therapeutics, and co-founded and serves as a consultant to 467
Inzen Therapeutics and Nevrox Limited. 468
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
24
References 469
1. Wu, F. et al. A new coronavirus associated with human respiratory disease in China. Nature 470
579, 265–269 (2020). 471
2. Zhou, P. et al. A pneumonia outbreak associated with a new coronavirus of probable bat origin. 472
Nature 579, 270–273 (2020). 473
3. Pillaiyar, T., Manickam, M., Namasivayam, V., Hayashi, Y. & Jung, S.-H. An Overview of Severe 474
Acute Respiratory Syndrome–Coronavirus (SARS-CoV) 3CL Protease Inhibitors: 475
Peptidomimetics and Small Molecule Chemotherapy. J. Med. Chem. 59, 6595–6628 (2016). 476
4. de Wit, E., van Doremalen, N., Falzarano, D. & Munster, V. J. SARS and MERS: recent insights 477
into emerging coronaviruses. Nat. Rev. Microbiol. 14, 523–534 (2016). 478
5. Huang, C., Wei, P., Fan, K., Liu, Y. & Lai, L. 3C-like Proteinase from SARS Coronavirus 479
Catalyzes Substrate Hydrolysis by a General Base Mechanism †. Biochemistry 43, 4568–4574 480
(2004). 481
6. Berry, M., Fielding, B. & Gamieldien, J. Human coronavirus OC43 3CL protease and the 482
potential of ML188 as a broad-spectrum lead compound: Homology modelling and molecular 483
dynamic studies. BMC Struct. Biol. 15, (2015). 484
7. Blanco, R., Carrasco, L. & Ventoso, I. Cell Killing by HIV-1 Protease. J. Biol. Chem. 278, 1086–485
1093 (2003). 486
8. Li, H. et al. Zika Virus Protease Cleavage of Host Protein Septin-2 Mediates Mitotic Defects in 487
Neural Progenitors. Neuron 101, 1089-1098.e4 (2019). 488
9. Chau, D. H. W. et al. Coxsackievirus B3 proteases 2A and 3C induce apoptotic cell death 489
through mitochondrial injury and cleavage of eIF4GI but not DAP5/p97/NAT1. Apoptosis 12, 490
513–524 (2007). 491
10. M’Barek, N. B., Audoly, G., Raoult, D. & Gluschankof, P. HIV-2 Protease resistance defined in 492
yeast cells. Retrovirology 3, 58 (2006). 493
11. Barco, A., Feduchi, E. & Carrasco, L. Poliovirus Protease 3Cpro Kills Cells by Apoptosis. 494
Virology 266, 352–360 (2000). 495
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
25
12. Li, M.-L. et al. The 3C protease activity of enterovirus 71 induces human neural cell apoptosis. 496
Virology 293, 386–395 (2002). 497
13. Jin, Z. et al. Structure of Mpro from SARS-CoV-2 and discovery of its inhibitors. Nature 582, 498
289–293 (2020). 499
14. Zhu, W. et al. Identification of SARS-CoV-2 3CL Protease Inhibitors by a Quantitative High-500
throughput Screening. bioRxiv 2020.07.17.207019 (2020) doi:10.1101/2020.07.17.207019. 501
15. CDC. Information for Laboratories about Coronavirus (COVID-19). Centers for Disease Control 502
and Prevention https://www.cdc.gov/coronavirus/2019-ncov/lab/lab-biosafety-guidelines.html 503
(2020). 504
16. Pyrc, K. et al. Culturing the Unculturable: Human Coronavirus HKU1 Infects, Replicates, and 505
Produces Progeny Virions in Human Ciliated Airway Epithelial Cell Cultures. J. Virol. 84, 11255–506
11263 (2010). 507
17. Sacco, M. D. et al. Structure and inhibition of the SARS-CoV-2 main protease reveals strategy 508
for developing dual inhibitors against Mpro and cathepsin L. bioRxiv 2020.07.27.223727 (2020) 509
doi:10.1101/2020.07.27.223727. 510
18. Sies, H. & Parnham, M. J. Potential therapeutic use of ebselen for COVID-19 and other 511
respiratory viral infections. Free Radic. Biol. Med. 156, 107–112 (2020). 512
19. Feoktistova, M., Geserick, P. & Leverkus, M. Crystal Violet Assay for Determining Viability of 513
Cultured Cells. Cold Spring Harb. Protoc. 2016, pdb.prot087379 (2016). 514
20. Ma, C. et al. Boceprevir, GC-376, and calpain inhibitors II, XII inhibit SARS-CoV-2 viral 515
replication by targeting the viral main protease. Cell Res. 1–15 (2020) doi:10.1038/s41422-020-516
0356-z. 517
21. Vuong, W. et al. Feline coronavirus drug inhibits the main protease of SARS-CoV-2 and blocks 518
virus replication. bioRxiv 2020.05.03.073080 (2020) doi:10.1101/2020.05.03.073080. 519
22. Froggatt, H. M., Heaton, B. E. & Heaton, N. S. Development of a fluorescence based, high-520
throughput SARS-CoV-2 3CLpro reporter assay. bioRxiv 2020.06.24.169565 (2020) 521
doi:10.1101/2020.06.24.169565. 522
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
26
23. Luan, X. et al. Structure Basis for Inhibition of SARS-CoV-2 by the Feline Drug GC376. bioRxiv 523
2020.06.07.138677 (2020) doi:10.1101/2020.06.07.138677. 524
24. Iketani, S. et al. Lead compounds for the development of SARS-CoV-2 3CL protease inhibitors. 525
bioRxiv 2020.08.03.235291 (2020) doi:10.1101/2020.08.03.235291. 526
25. Yang, S. et al. Synthesis, Crystal Structure, Structure−Activity Relationships, and Antiviral 527
Activity of a Potent SARS Coronavirus 3CL Protease Inhibitor. J. Med. Chem. 49, 4971–4980 528
(2006). 529
26. Dai, W. et al. Structure-based design of antiviral drug candidates targeting the SARS-CoV-2 530
main protease. Science 368, 1331–1335 (2020). 531
27. Ghosh, A. K. et al. Design, synthesis and antiviral efficacy of a series of potent chloropyridyl 532
ester-derived SARS-CoV 3CLpro inhibitors. Bioorg. Med. Chem. Lett. 18, 5684–5688 (2008). 533
28. Węglarz-Tomczak, E., Tomczak, J. M., Talma, M. & Brul, S. Ebselen as a highly active inhibitor 534
of PLProCoV2. bioRxiv 2020.05.17.100768 (2020) doi:10.1101/2020.05.17.100768. 535
29. Seale, L. A., Torres, D. J., Berry, M. J. & Pitts, M. W. A role for selenium-dependent GPX1 in 536
SARS-CoV-2 virulence. Am. J. Clin. Nutr. doi:10.1093/ajcn/nqaa177. 537
30. Kim, Y. et al. Reversal of the Progression of Fatal Coronavirus Infection in Cats by a Broad-538
Spectrum Coronavirus Protease Inhibitor. PLOS Pathog. 12, e1005531 (2016). 539
31. Zhang, J.-H., Chung, T. D. Y. & Oldenburg, K. R. A Simple Statistical Parameter for Use in 540
Evaluation and Validation of High Throughput Screening Assays. J. Biomol. Screen. 4, 67–73 541
(1999). 542
32. Birmingham, A. et al. Statistical Methods for Analysis of High-Throughput RNA Interference 543
Screens. Nat. Methods 6, 569–575 (2009). 544
33. Kim, Y. et al. Broad-Spectrum Antivirals against 3C or 3C-Like Proteases of Picornaviruses, 545
Noroviruses, and Coronaviruses. J. Virol. 86, 11754–11762 (2012). 546
34. Blanchard, J. E. et al. High-throughput screening identifies inhibitors of the SARS coronavirus 547
main proteinase. Chem. Biol. 11, 1445–1453 (2004). 548
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
27
35. Klotz, L. C. & Sylvester, E. J. The Consequences of a Lab Escape of a Potential Pandemic 549
Pathogen. Front. Public Health 2, (2014). 550
36. Coelho, A. C. & García Díez, J. Biological Risks and Laboratory-Acquired Infections: A Reality 551
That Cannot be Ignored in Health Biotechnology. Front. Bioeng. Biotechnol. 3, (2015). 552
37. Tan, C. W. et al. A SARS-CoV-2 surrogate virus neutralization test based on antibody-mediated 553
blockage of ACE2–spike protein–protein interaction. Nat. Biotechnol. (2020) 554
doi:10.1038/s41587-020-0631-z. 555
38. Case, J. B. et al. Neutralizing Antibody and Soluble ACE2 Inhibition of a Replication-Competent 556
VSV-SARS-CoV-2 and a Clinical Isolate of SARS-CoV-2. Cell Host Microbe 557
doi:10.1016/j.chom.2020.06.021. 558
39. Dieterle, M. E. et al. A Replication-Competent Vesicular Stomatitis Virus for Studies of SARS-559
CoV-2 Spike-Mediated Cell Entry and Its Inhibition. Cell Host Microbe 560
doi:10.1016/j.chom.2020.06.020. 561
40. Iglewicz, B. & Hoaglin, D. C. How to detect and handle outliers. (ASQC Quality Press, 1993). 562
41. Metzger, A., Melzig, L., Despotopoulou, C. & Knochel, P. Pd-Catalyzed Cross-Coupling of 563
Functionalized Organozinc Reagents with Thiomethyl-Substituted Heterocycles. Org. Lett. 11, 564
4228–4231 (2009). 565
42. Zambaldo, C. et al. 2-Sulfonylpyridines as Tunable, Cysteine-Reactive Electrophiles. J. Am. 566
Chem. Soc. 142, 8972–8979 (2020). 567
43. Mou, K. et al. Novel CADD-based peptidyl vinyl ester derivatives as potential proteasome 568
inhibitors. Bioorg. Med. Chem. Lett. 18, 2198–2202 (2008). 569
44. Amblard, F. et al. Synthesis and antiviral evaluation of novel peptidomimetics as norovirus 570
protease inhibitors. Bioorg. Med. Chem. Lett. 28, 2165–2170 (2018). 571
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
28
Fig. 1. Expression of SARS-CoV-2 3CLpro in HEK293T cells results in toxicity that can be 572
rescued by GC376. a. SARS-CoV-2 3CL toxicity is dependent on protease activity and can be 573
visualized with crystal violet staining. b. Quantification of crystal violet staining in a. c. Treatment of 574
SARS-CoV-2 3CLpro expressing cells with protease inhibitor GC376 results in rescue of cytotoxicity. 575
d. Quantification of c. Data are shown as mean ± s.d. for four technical replicates. 576
577
a b
c d
EYFP
SARS-CoV-2 3CLpro
SARS-CoV-2 3CLpro C145A
SARS-CoV-2 3CLpro + Vehicle
SARS-CoV-2 3CLpro + 50 µM GC376
EYFP
SARS-CoV-2
3CLpro
SARS-CoV-2
3CLpro
C14
5A0.0
0.2
0.4
0.6
Abs
orba
nce
Vehicle 50µM GC3760.0
0.2
0.4
0.6
Abs
orba
nce
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
29
Fig. 2. Dose response curves for SARS-CoV-2 3CLpro can be conducted with transfection-578
based assays. a-c. SARS-CoV-2 3CLpro can be inhibited by known 3CLpro inhibitors GC376, 579
compound 4, and 11a. The toxicity of each compound was determined by treating EYFP-transfected 580
cells with indicated concentrations of drug and is reported as Cell Viability. d. Chemical structures for 581
each of the compounds tested. EC50 values are displayed as best-fit value alongside 95% 582
confidence interval. CC50 values are displayed as best-fit value. Data are shown as mean ± s.d. for 583
four technical replicates. 584
585
-6 -4 -2 0 2 40
1
2
3
4
0
50
100
150
log[GC376], µM
Rel
ativ
e G
row
th
EC50 = 3.30 µM95% CI: [1.81 - 6.04]
CC50 = >100 µM
Cell V
iability (%)
a b
c
-6 -4 -2 0 2 40
1
2
3
4
0
50
100
150
log[Compound 4], µM
Rel
ativ
e G
row
th
EC50 = 0.98 µM95% CI: [0.48 - 2.01]
CC50 = 81 µMRelative Growth
Cell Viability
Cell V
iability (%)
GC376 Compound 4
11a
d
-6 -4 -2 0 2 40
1
2
3
4
0
50
100
150
log[11a], µM
Rel
ativ
e G
row
th
EC50 = 6.89 µM95% CI: [3.80 - 12.59]
CC50 = 48 µM
Cell V
iability (%)
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
30
Fig. 3. The activity of GC376, compound 4, and 11a show variable effectiveness and potency 586
against the coronavirus 3CL proteases from SARS-CoV, MERS-CoV, Bat-CoV-HKU9, HCoV-587
NL63, and IBV. EC50 values are displayed as best-fit value alongside 95% confidence interval. CC50 588
values are displayed as best-fit value. Data are shown as mean ± s.d. for three or four technical 589
replicates. 590
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
31
591
GC376 Compound 4 11aSA
RS-
CoV
3CLp
roM
ERS-
CoV
3CLp
roB
at-C
oV-H
KU
9 3C
Lpro
HC
oV-N
L63
3CLp
roIB
V 3C
Lpro
-6 -4 -2 0 2 40
1
2
3
0
50
100
150
log[11a], µM
Rel
ativ
e G
row
th
Cell V
iability (%)
-6 -4 -2 0 2 40
1
2
3
0
50
100
150
log[11a], µM
Rel
ativ
e G
row
th
Cell V
iability (%)
-4 -2 0 2 40
1
2
3
4
0
50
100
150
log[GC376], µM
Rel
ativ
e G
row
thEC50 = 5.83 µM
95% CI: [3.58 - 9.38] Cell V
iability (%)
-6 -4 -2 0 2 40
1
2
3
4
5
0
50
100
150
log[Compound 4], µM
Rel
ativ
e G
row
th
EC50 = 3.17 µM95% CI: [2.03 - 4.99] C
ell Viability (%
)
-6 -4 -2 0 2 40
1
2
3
0
50
100
150
log[11a], µM
Rel
ativ
e G
row
th
EC50 = 5.38 µM95% CI: [1.63 - 32.5]
Cell V
iability (%)
-4 -2 0 2 40
1
2
3
0
50
100
150
log[GC376], µM
Rel
ativ
e G
row
th
EC50 = 7.44 µM95% CI: [4.01 - 14.04] C
ell Viability (%
)
-6 -4 -2 0 2 40
1
2
3
4
0
50
100
150
log[Compound 4], µM
Rel
ativ
e G
row
th
EC50 = 1.40 µM95% CI: [0.78 - 2.48]
Cell V
iability (%)
-4 -2 0 2 40
1
2
3
0
50
100
150
log[GC376], µM
Rel
ativ
e G
row
th
EC50 = 11.07 µM95% CI: [6.07 - 20.9] C
ell Viability (%
)
-6 -4 -2 0 2 40
1
2
3
0
50
100
150
log[Compound 4], µM
Rel
ativ
e G
row
th
EC50 = 4.11 µM95% CI: [2.18 - 7.86]
Cell V
iability (%)
-4 -2 0 2 40
1
2
3
4
0
50
100
150
log[GC376], µM
Rel
ativ
e G
row
th
EC50 = >50 µM
Cell V
iability (%)
-6 -4 -2 0 2 40
1
2
3
4
0
50
100
150
log[Compound 4], µM
Rel
ativ
e G
row
th
EC50 = 4.92 µM95% CI: [3.53 - 6.85]
Cell V
iability (%)
-6 -4 -2 0 2 40
1
2
3
0
50
100
150
log[11a], µM
Rel
ativ
e G
row
th
EC50 = >20 µM
Cell V
iability (%)
-4 -2 0 2 40
1
2
3
0
50
100
150
log[GC376], µM
Rel
ativ
e G
row
th
EC50 = 0.58 µM95% CI: [0.27 - 1.21]
Cell V
iability (%)
-6 -4 -2 0 2 40
1
2
3
4
0
50
100
150
log[Compound 4], µM
Rel
ativ
e G
row
th
EC50 = 0.058 µM95% CI: [0.028 - 0.1]
Cell V
iability (%)
-6 -4 -2 0 2 40
1
2
3
0
50
100
150
log[11a], µM
Rel
ativ
e G
row
th
EC50 = >20 µM
Cell V
iability (%)
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
32
Fig. 4. Small-scale drug screen and structure-activity profiling at 10 µM identify two 592
compounds, GC373 and GRL-0496, with activity against the SARS-CoV-2 3CLpro. a. 593
Identification of hits from the drug screen and structure-activity profiling. Positive control compounds 594
were included in each plate and are highlighted. b. Compounds with structural similarity to known 595
inhibitors. Compounds in bold are molecules that show activity against the SARS-CoV-2 3CLpro at 596
10 µM. c. Dose-response profiling and cytotoxicity determination of GRL-0496 against the SARS-597
CoV-2 3CLpro. d. Live virus testing of GRL-0496 against SARS-CoV-2. 598
EC50 values are displayed as best-fit value alongside 95% confidence interval. The live virus assay 599
was conducted with two biological replicates, each with three technical replicates and the EC50 value 600
was derived from all replicates. CC50 values are displayed as best-fit value. Data are shown as mean 601
± s.d. for three or four technical replicates. 602
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
33
603
a
MAC-5576 BTB07408 BTB07417GRL-0496
Compound 4 AZVIII-34D
GC376 AZVIII-41A GC373
b
c
-3 -2 -1 0 1 2
0
50
100
log[GRL-0496], µM
% S
AR
S-C
oV-2
Inhi
bitio
n
EC50 = 9.12 µM95% CI: [7.00 - 11.88]
d
-6 -4 -2 0 2 4
1
2
0
50
100
150
log[GRL-0496], µM
Rel
ativ
e G
row
th
EC50 = 5.05 µM95% CI: [2.99 - 8.56]
CC50 = 81 uM Cell V
iability (%)
-20
0
20
40
60
Rob
ust Z
Sco
re
Plate 1 Plate 2 Plate 3 Plate 4
GRL-0496
GC376 - 50 µMCompound 4 - 20 µM
GC373 Controls
Compound 4 - 10 µM
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint
34
Table 1. Comparison of literature reported live virus based EC50 values compared to values 604
generated during this study. CPE = Cytopathic effect. 605
Protease Drug Calculated
EC50 (µM)
Literature
Reported
Value (µM)
Method Cell
Line
Citation
SARS-CoV-2
3CLpro
GC376 3.3 3.37 CPE Vero 76 Ma et. al.
0.9 Plaque
Assay
Vero E6 Vuong et. al.
0.18 qPCR Vero E6 Luan et. al.
2.2 qPCR Vero E6 Froggatt et. al.
4.48 CPE Vero E6 Iketani et. al.
SARS-CoV-2
3CLpro
11a 6.89 2.06 CPE Vero E6 This study
0.53 Plaque
Assay
Vero E6 Dai et. al.
SARS-CoV-2
3CLpro
Compound
4
0.98 3.023 CPE Vero E6 Iketani et. al.
SARS-CoV-2
3CLpro
GRL-0496 5.05 9.12 CPE Vero E6 This study
SARS-CoV
3CLpro
GRL-0496 7.84 6.9 CPE Vero E6 Ghosh et. al.
SARS-CoV-2
3CLpro
GC373 2.8 1.5 Plaque
Assay
Vero E6 Vuong et. al.
4.83 CPE Vero E6 This Study
606
.CC-BY-NC-ND 4.0 International licenseavailable under a(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 29, 2020. ; https://doi.org/10.1101/2020.08.29.272864doi: bioRxiv preprint